Salvato in:
Dettagli Bibliografici
Autori principali: Nikitin, Filipp, Dunn, Ian, Koes, David Ryan, Isayev, Olexandr
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2505.00169
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909611808784384
author Nikitin, Filipp
Dunn, Ian
Koes, David Ryan
Isayev, Olexandr
author_facet Nikitin, Filipp
Dunn, Ian
Koes, David Ryan
Isayev, Olexandr
contents Deep generative models have shown significant promise in generating valid 3D molecular structures, with the GEOM-Drugs dataset serving as a key benchmark. However, current evaluation protocols suffer from critical flaws, including incorrect valency definitions, bugs in bond order calculations, and reliance on force fields inconsistent with the reference data. In this work, we revisit GEOM-Drugs and propose a corrected evaluation framework: we identify and fix issues in data preprocessing, construct chemically accurate valency tables, and introduce a GFN2-xTB-based geometry and energy benchmark. We retrain and re-evaluate several leading models under this framework, providing updated performance metrics and practical recommendations for future benchmarking. Our results underscore the need for chemically rigorous evaluation practices in 3D molecular generation. Our recommended evaluation methods and GEOM-Drugs processing scripts are available at https://github.com/isayevlab/geom-drugs-3dgen-evaluation.
format Preprint
id arxiv_https___arxiv_org_abs_2505_00169
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle GEOM-Drugs Revisited: Toward More Chemically Accurate Benchmarks for 3D Molecule Generation
Nikitin, Filipp
Dunn, Ian
Koes, David Ryan
Isayev, Olexandr
Machine Learning
Artificial Intelligence
Deep generative models have shown significant promise in generating valid 3D molecular structures, with the GEOM-Drugs dataset serving as a key benchmark. However, current evaluation protocols suffer from critical flaws, including incorrect valency definitions, bugs in bond order calculations, and reliance on force fields inconsistent with the reference data. In this work, we revisit GEOM-Drugs and propose a corrected evaluation framework: we identify and fix issues in data preprocessing, construct chemically accurate valency tables, and introduce a GFN2-xTB-based geometry and energy benchmark. We retrain and re-evaluate several leading models under this framework, providing updated performance metrics and practical recommendations for future benchmarking. Our results underscore the need for chemically rigorous evaluation practices in 3D molecular generation. Our recommended evaluation methods and GEOM-Drugs processing scripts are available at https://github.com/isayevlab/geom-drugs-3dgen-evaluation.
title GEOM-Drugs Revisited: Toward More Chemically Accurate Benchmarks for 3D Molecule Generation
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2505.00169